Clonal heterogeneity associated with acquired resistance presents a critical therapeutic challenge. Whole-exome sequencing of paired tumor biopsies and targeted sequencing of cell-free DNA (cfDNA) from patients with BRAFV600E colorectal cancer receiving BRAF inhibitor combinations identified 14 distinct alterations in MAPK pathway components driving acquired resistance, with as many as eight alterations in a single patient. We developed a pooled clone system to study clonal outgrowth during acquired resistance, in vitro and in vivo. In vitro, the dynamics of individual resistant clones could be monitored in real time in cfDNA isolated from culture media during therapy. Outgrowth of multiple resistant clones was observed during therapy with BRAF, EGFR, and MEK inhibitor combinations. However, ERK inhibition, particularly in combination with BRAF and EGFR inhibition, markedly abrogated clonal outgrowth in vitro and in vivo. Thus, convergent, up-front therapy may suppress outgrowth of heterogeneous clones harboring clinically observed resistance alterations, which may improve clinical outcome.
Significance: We observed heterogeneous, recurrent alterations in the MAPK pathway as key drivers of acquired resistance in BRAFV600E colorectal cancer, with multiple concurrent resistance alterations detectable in individual patients. Using a novel pooled clone system, we identify convergent up-front therapeutic strategies capable of intercepting multiple resistance mechanisms as potential approaches to suppress emergence of acquired resistance. Cancer Discov; 8(4); 417–27. ©2018 AACR.
See related commentary by Janku, p. 389.
See related article by Corcoran et al., p. 428.
This article is highlighted in the In This Issue feature, p. 371
Approximately 10% of patients with colorectal cancer harbor BRAFV600E mutations, which confer poor prognosis in the metastatic setting and lead to constitutive activation of the MAPK signaling pathway (1–3). Normally, BRAF and the other RAF kinases (ARAF and CRAF) are activated by RAS family members (KRAS, NRAS, or HRAS) in response to signals from receptor tyrosine kinases (RTK). Once activated, BRAF and other RAF kinases phosphorylate and activate MEK kinases (MEK1 and MEK2), which in turn phosphorylate and activate ERK kinases (ERK1 and ERK2). However, BRAFV600E leads to constitutive BRAF kinase activity and downstream MAPK signaling through MEK and ERK.
Although melanomas harboring BRAFV600E mutations have shown dramatic response rates of >50% to single-agent BRAF inhibitors (BRAFi), such as vemurafenib and dabrafenib, efficacy has been disappointing in BRAFV600E colorectal cancer, with response rates of only 5% (4–6). Preclinical studies have suggested that BRAFi resistance in colorectal cancer is driven by reactivation of MAPK signaling through loss of negative feedback signals, leading to activation of RAS by RTKs, such as EGFR (7–10). Based on these findings, recent clinical trials have evaluated combinations of BRAFi with EGFRi, MEKi, or the triple combination of BRAFi, EGFRi, and MEKi in patients with BRAFV600E colorectal cancer, with the goal of producing more sustained MAPK inhibition and improved efficacy. Although BRAFi combinations have led to modestly improved response rates in colorectal cancer, these responses are not durable, and clinical benefit is limited by the rapid emergence of acquired resistance (11–16).
Previously, our group and others published the first clinical mechanisms of acquired resistance to BRAFi/EGFRi and BRAFi/MEKi combinations in BRAFV600E colorectal cancer (17–19). Interestingly, we observed an array of resistance alterations affecting MAPK signaling components, all converging to reactivate the MAPK pathway despite the presence of multiple inhibitors. These resistance alterations included KRAS mutation and amplification, BRAFV600E amplification, MEK mutation, and amplification of RTKs such as MET. These findings underscore the critical dependence of BRAFV600E colorectal cancer on the MAPK pathway, but also create a challenging therapeutic scenario in which many distinct resistance mechanisms may need to be targeted in different patients.
This challenge is further complicated by the potential for tumor heterogeneity associated with acquired resistance. Although the role of tumor heterogeneity in BRAFV600E colorectal cancer has yet to be clearly defined, studies of other molecular subtypes of colorectal cancer and other tumor types have demonstrated the potential for multiple resistance alterations to emerge in distinct tumor subclones in an individual patient under the selective pressure of targeted therapy (20, 21). The heterogeneity of acquired resistance is thought to be dynamically shaped through selection and expansion of tumor subclones with preexisting alterations that sustain growth under targeted pressure and can result in the emergence of distinct resistant subclones in different metastatic lesions, or in the same lesion (22–24). Thus, to overcome resistance in BRAFV600E colorectal cancer, there is an urgent need for a strategy capable of surmounting multiple heterogeneous resistance mechanisms, both among patients, and among different tumor subclones in the same patient.
In this study, we explored the role of heterogeneity in acquired resistance in BRAFV600E colorectal cancer and evaluated whether targeting a convergent signaling node could intercept multiple resistance mechanisms and prevent outgrowth of heterogeneous resistant clones. We observed recurrent resistance alterations in the MAPK pathway in patients with BRAFV600E colorectal cancer, with as many as eight concurrent alterations detected in a single patient. Through a novel pooled clone model, we observed that convergent, up-front therapy can suppress outgrowth of multiple clones harboring a spectrum of clinically observed resistance alterations, suggesting a promising approach for future clinical trials.
Heterogeneity and Acquired Resistance in BRAFV600E Colorectal Cancer
We evaluated postprogression tumor biopsies and cell-free DNA (cfDNA) from patients with BRAFV600E colorectal cancer achieving response or stable disease on BRAFi combinations to study the emergence of acquired resistance. Patients 1 to 3 (Fig. 1A–C; Supplementary Table S1) received combined BRAFi/EGFRi (dabrafenib + panitumumab), BRAFi/EGFRi/MEKi (dabrafenib + panitumumab + trametinib), and BRAFi/EGFRi/PI3Ki (encorafenib + cetuximab + alpelisib), respectively. Compared with pretreatment specimens, we observed the persistent presence in postprogression tumor or cfDNA of the original BRAFV600E mutation and patient-specific founder mutations (in TP53, APC, or SMAD4) present in the pretreatment tumor specimen. However, the emergence of one or more RAS mutations (KRAS or NRAS), not detectable in the pretreatment specimen, was observed in each patient upon disease progression, including KRASG12D, KRASG12S, NRASQ61K, and NRASQ61R. Similarly, in our co-submitted clinical study, we observed emergence of one or more RAS mutations in cfDNA at disease progression in 14 of 29 (48%) patients with BRAFV600E colorectal cancer who responded or achieved stable disease on combined BRAFi/EGFRi/MEKi therapy with dabrafenib, panitumumab, and trametinib (16). Collectively, these findings corroborate initial data from our group and others, which showed that RAS mutations and other alterations leading to MAPK reactivation are key mechanisms of acquired resistance in BRAFV600E colorectal cancer treated with two-drug BRAFi combinations, and suggest that these same mechanisms can drive clinical acquired resistance to three-drug BRAFi combinations.
Interestingly, patient 3 developed two emergent RAS alterations—KRASG12S and NRASQ61R—in his postprogression cfDNA following 17 months of therapy with combined BRAFi/EGFRi/PI3Ki (Fig. 1C). In a biopsy of a progressing liver metastasis (Fig. 1C, bottom; Supplementary Table S1) obtained upon treatment failure, the same KRASG12S mutation was detected, but the NRASQ61R mutation was not, suggesting that this alteration may have emerged in a different metastatic lesion. This potential for tumor heterogeneity in the setting of acquired resistance is also evident in our co-submitted clinical study, in which 6 of 14 (42%) patients with emergent RAS mutations upon disease progression following treatment with dabrafenib, panitumumab, and trametinib developed more than one RAS mutation detectable in cfDNA, with as many as four RAS mutations observed in an individual patient (16).
Similarly, striking heterogeneity associated with acquired resistance was observed in patient 4, following progression on dabrafenib, panitumumab, and trametinib (Fig. 1D; Supplementary Table S1). In postprogression cfDNA, the original BRAFV600E mutation as well as three additional founder mutations (APC, SMAD4, and PTEN) present prior to treatment remained present. Remarkably, however, this patient harbored eight coexisting emergent resistance alterations detectable in postprogression cfDNA, including BRAFV600E amplification, three KRAS mutations, an NRAS mutation, two MEK1 (MAP2K1) mutations, and an MEK2 (MAP2K2) mutation, all of which converge on MAPK reactivation.
Collectively, these cases demonstrate that a diverse array of potential resistance mechanisms can arise among patients, but also among different tumor subclones in the same patient, posing a significant challenge to the development of therapies to overcome resistance. However, all of these resistance mechanisms converge upon reactivation of the MAPK pathway, specifically on activation of ERK. This common thread creates a potential therapeutic opportunity to devise a single strategy capable of intercepting and overcoming multiple heterogeneous resistance mechanisms simultaneously.
Convergent Inhibition of MAPK Signaling to Overcome Resistance
Previously, we reported that because multiple clinical resistance mechanisms in BRAFV600E colorectal cancer appear to converge on ERK reactivation, ERK inhibition may represent a promising strategy to overcome resistance. However, the efficacy of ERK inhibition, alone or in combination with other MAPK inhibitors, has not been carefully assessed across the broad spectrum of clinically observed resistance mechanisms.
We engineered seven distinct mutations in KRAS, NRAS, MEK1, and MEK2, encompassing all amino acid positions mutated in the context of clinical acquired resistance in our current and prior study, in the BRAFV600E colorectal cancer cell line VACO432, which is sensitive to BRAF inhibitor combinations (Supplementary Fig. S1A and S1B). Each resistant cell line was cultured in the presence of eight different drug combinations, including combinations currently in clinical trials, as well as ERKi alone or in combination with other MAPK pathway inhibitors. These agents included the BRAFi dabrafenib, the MEKi trametinib, the EGFRi panitumumab, and the ERKi VX-11e. Individually, each mutation led to a marked shift in IC50 (Supplementary Table S2) to BRAFi/EGFRi or BRAFi/MEKi combinations in a short-term viability assay (Fig. 2A) compared with empty-vector control cells. However, no clear shift in IC50 was noted to ERKi alone or in combinations with other MAPK pathway inhibitors. Interestingly, no IC50 shift was noted with BRAFi/MEKi/EGFRi, even though many of these resistance mechanisms were observed clinically in patients treated with this therapy, suggesting that short-term viability assays may not be sufficient to discern the role of each mutation in driving resistance.
Thus, we next evaluated the ability of these treatment combinations to suppress long-term viability over the course of 3 weeks. Again, all clones drove rapid resistance to BRAFi/EGFRi or BRAFi/MEKi combinations compared with control cells (Fig. 2B). Notably, although BRAFi/MEKi/EGFRi was able to suppress the growth of each clone throughout the first week of treatment, clear outgrowth was observed relative to empty-vector control by the end of the 3-week period. These results are consistent with the clinical observation that these mutations drive acquired resistance to this triplet therapy, but also suggest that resistance may be delayed and potentially involve adaptive changes. Long-term viability also revealed differences in the efficacy of ERKi alone or in combination. Although ERKi alone suppressed the outgrowth of each model overall relative to the non-ERKi combinations tested, some degree of resistance was observed relative to empty-vector control. Although the addition of EGFRi did not noticeably alter efficacy, the addition of BRAFi or MEKi appeared to improve efficacy across the spectrum of resistant models. However, BRAFi/ERKi/EGFRi triple blockade was able to completely abrogate growth of each model relative to control.
To better understand the effects of each therapy, we evaluated the signaling consequences in each resistant model. Each resistant model exhibited increased basal levels of phosphorylated ERK (pERK) and phosphorylation of the downstream ERK target RSK (pRSK; Supplementary Fig. S2). Each mutation also led to a decreased ability of BRAFi/EGFRi or BRAFi/MEKi, and to a lesser extent BRAFi/MEKi/EGFRi, to inhibit pERK and pRSK relative to empty-vector control. In sharp contrast, no marked shift in the ability of ERKi-based treatments to inhibit pRSK was observed. Of note, because ERK inhibition can lead to feedback induction of pERK, pERK is not a reliable marker of MAPK pathway inhibition in the presence of ERKi, whereas pRSK provides a readout of ERK activity. Thus, improved suppression of MAPK signaling by ERKi in the presence of these resistance alterations likely underlies the ability of these therapies to suppress the long-term viability of these models.
To further dissect the differential signaling effects of each key therapy, we compared their ability to suppress the ERK transcriptional targets DUSP6 and SPRY2 as a readout of MAPK pathway output over time (Fig. 2C; Supplementary Fig. S3). Although BRAFi/EGFRi and BRAFi/MEKi were able to suppress DUSP6 and SPRY2 transcript levels in empty-vector control cells throughout the full treatment period, the ability of these combinations to suppress transcript levels in the presence of each resistance mutation was markedly abrogated, and even a modest induction of transcript levels was observed in the KRASG12D clone. Consistent with the delayed outgrowth of resistant clones seen in long-term viability assays with BRAFi/MEKi/EGFRi, an initial reduction in transcript levels was observed across all models, but transcript levels rebounded rapidly by 36 to 48 hours, suggesting that the ability of this treatment to suppress MAPK signaling is transient in the presence of these mutations (Fig. 2C; Supplementary Fig. S4). By contrast, ERKi was able to suppress transcript levels in all models, but to a lesser degree than in control cells. However, BRAFi/ERKi and to a more complete degree BRAFi/ERKi/EGFRi were able to suppress transcript levels across all models to a similar degree as in control cells, perhaps explaining the observed differences in long-term growth suppression. Taken together, these results indicate that convergent targeting of ERK has the potential to overcome the spectrum of clinically observed resistance mutations, but that ERKi may more effectively suppress MAPK signaling and outgrowth of resistant clones when combined with BRAFi and/or EGFRi.
Real-Time Modeling of Heterogeneity and Acquired Resistance
Acquired resistance is thought to arise primarily from rare preexisting subclones harboring resistance alterations that emerge under the selective pressure of therapy (22, 25, 26). Therefore, rather than waiting for acquired resistance to emerge before applying a therapy designed to overcome resistance, a potentially promising alterative would be to apply that therapy in the up-front setting to suppress and ideally eradicate preexisting resistant clones while they constitute a low-frequency subpopulation. Thus, to study clonal outgrowth during treatment in a more physiologically relevant setting, we developed a pooled clone heterogeneity model. In this system, each resistant clone was pooled at an abundance of 1% in a background of sensitive empty-vector control cells (Fig. 3A).
To investigate the growth characteristics of the clonal pool upon exposure to each therapy, the total number of viable cells was counted at intervals during treatment (Fig. 3B). In parallel, the abundance of each clone at baseline and at the end of treatment was determined by droplet-digital PCR (ddPCR) using genomic DNA isolated from pelleted cells (Fig. 3C, pie charts).
Interestingly, we found that cfDNA shed from cells in tissue culture could be isolated from cell culture media supernatant and analyzed. Moreover, we observed that cell culture media–derived cfDNA recapitulated the mutant allele percentage measured in genomic DNA derived from harvested cells (Supplementary Fig. S5A), and that the half-life of cfDNA was approximately 2 hours (Supplementary Fig. S5B). These findings suggest that cfDNA isolated from cell culture media can provide a real-time assessment of the clonal abundance within the cell population. This insight allowed us to monitor clonal dynamics within the pooled population in real time by isolating cfDNA from cell culture media every 3 to 4 days during therapy for analyses by ddPCR (Fig. 3C, graphs) to better understand the kinetics of clonal outgrowth.
Relative to untreated cells, BRAFi/EGFRi and BRAFi/MEKi led to only a slight delay in growth of the clonal pool and showed massive expansion in the abundance of resistant clones (Fig. 3B and C), suggesting that overall expansion of the population was driven primarily by outgrowth of these resistant clones. Interestingly, BRAFi/MEKi/EGFRi treatment led to an initial reduction in overall cell number followed by a rapid increase in cell number and robust expansion in the abundance of resistant clones. This pattern is consistent with initial efficacy against the control population followed by a delayed expansion of resistant clones. Indeed, the real-time kinetics afforded by serial cfDNA analysis from the culture media show that outgrowth of resistant clones is initially suppressed, but that expansion occurs following the first week of treatment, suggesting that adaptive changes over time may enhance the ability of these resistant clones to proliferate in the presence of BRAFi/MEKi/EGFRi, consistent with the long-term viability data (Fig. 2B) and the transient pathway inhibition observed in Fig. 2C.
In contrast, ERKi alone and in doublet combinations led to a largely cytostatic effect on overall cell number throughout the treatment period, but abrogated the expansion of resistant clones (Fig. 3B and C). Indeed, the mild increase in abundance of each resistant clone in the pool was equal to or less than the increase seen in the vehicle-treated population, which may represent a slight basal proliferative advantage. Strikingly, BRAFi/ERKi/EGFRi led to a marked reduction in overall cell number, indicative of cell death, and completely prevented outgrowth of all resistant clones relative to pretreatment clonal percentage, suggesting that the selective advantage imparted by these mutations was completely nullified in the presence of this therapy. Indeed, by combining overall cell number (Fig. 3B) and relative clonal percentages (Fig. 3C) throughout therapy, we can interpolate absolute clonal abundance throughout the treatment period (Fig. 3D). These data show that only BRAFi/ERKi and BRAFi/ERKi/EGFRi prevent the outgrowth of resistant clones over the entire treatment period, with only BRAFi/ERKi/EGFRi leading to an overall reduction in the absolute abundance of each clone, indicative of a cytotoxic effect. Interestingly, for all treatment groups, control cells constituted some percentage of the final cell population, suggesting that these cells were not eradicated by treatment, but rather that a subset of these cells persisted, perhaps due to adaptive changes. Collectively, these findings highlight the potential utility of in vitro cfDNA analysis to model clonal evolution in real time and suggest that up-front treatment with ERKi in combination with BRAFi or BRAFi/EGFRi (as opposed to monotherapy) may be a promising strategy to suppress resistant outgrowth in BRAFV600E colorectal cancer.
Suppression of Clonal Outgrowth In Vivo
Next, we tested whether convergent inhibition was effective in vivo by growing these same clonal pools as xenografts. Tumor-bearing mice were treated with the six most clinically relevant drug combinations, and treatment was well tolerated (Supplementary Fig. S6). BVD-523 (ulixertinib), the clinical analogue of Vx-11e, was used for these studies. After the treatment period, tumors were harvested and genomic DNA was isolated and analyzed by ddPCR to determine clonal abundance relative to baseline tumors harvested at the start of treatment. BRAFi/EGFRi and BRAFi/MEKi treatment led to a slight initial delay in tumor growth, relative to vehicle treatment, followed by rapid outgrowth (Fig. 4A and B). Interestingly, BRAFi/MEKi/EGFRi, ERKi alone, or BRAFi/ERKi produced a more substantial reduction in tumor growth relative to vehicle, but tumors still increased in size overall, and no significant difference was observed among these treatment arms. However, BRAFi/ERKi/EGFRi demonstrated significantly increased antitumor effect, leading to substantial tumor regression in all treated animals. Marked outgrowth of resistant clones was observed in tumors treated with BRAFi/EGFRi, BRAFi/MEKi, and BRAFi/MEKi/EGFRi, relative to baseline tumors and vehicle-treated controls (Fig. 4C). In contrast, ERKi alone and BRAFi/ERKi repressed the outgrowth of resistant clones overall, with only KRASG12D and KRASQ61H showing modest increases in relative abundance. However, clonal outgrowth was completely suppressed in BRAFi/ERKi/EGFRi-treated mice, with all resistant clones remaining at levels equivalent to baseline tumors. These data suggest that up-front convergent therapy may represent a promising strategy to suppress the outgrowth of multiple heterogeneous resistance mechanisms in BRAFV600E colorectal cancer.
Although BRAFi combinations have led to important increases in response rates for patients with BRAFV600E colorectal cancer in recent clinical trials, the rapid emergence of acquired resistance represents a major barrier to clinical benefit, limiting median progression-free survival to only ∼4 months. In this study, along with the 29 patients analyzed in our co-submitted clinical study (16), we show recurrent MAPK pathway alterations are a primary mode of acquired resistance in patients with BRAFV600E colorectal cancer achieving initial response or stable disease. These data are consistent with initial smaller studies from our group and others of patients treated with BRAFi doublet combinations (17–19, 27) and confirm that MAPK alterations are also a major driver of acquired resistance to newer triplet BRAFi combinations. We also observed the potential for profound tumor heterogeneity associated with acquired resistance, with 2 of 4 patients in this study and 6 of 14 patients with detectable resistance alterations in our co-submitted clinical study harboring more than one resistance mechanism (16), with as many as eight resistance alterations observed in a single patient (Fig. 1D). These findings highlight the need for a strategy capable of simultaneously overcoming a diverse array of MAPK resistance alterations to combat acquired resistance.
Because all of the clinical resistance alterations observed thus far in BRAFV600E colorectal cancer converge on reactivation of MAPK signaling through activation of ERK, this provides a unique opportunity to target a common signaling node in order to intercept multiple upstream resistance mechanisms. Indeed, the potential promise of targeting convergent evolution in the setting of acquired resistance has been suggested by several preclinical studies in other molecularly defined cancer subtypes (28–30). Similarly, our data indicate that ERKi may represent a key component of strategies to intercept heterogeneous resistance alterations. Alternative strategies might involve agents such as pan-RAF or RAF dimer inhibitors (27). Interestingly, most of the clinically observed resistance mechanisms also converge upon MEK, yet upstream alterations such as RAS mutation are still observed with MEKi combinations (BRAFi/MEKi and BRAFi/MEKi/EGFRi). This observation is consistent with previous work by our group and others demonstrating that MEKi are vulnerable to resistance mechanisms that increase upstream signaling flux, leading to MEK hyperactivation and a decreased ability of MEKi to suppress signaling throughput to ERK (7, 31). Conversely, our data suggest that ERKi are able to overcome this vulnerability and can maintain pathway suppression in the setting of increased upstream pathway activation, thus constituting a more effective backbone for convergent targeting.
However, our data also suggest that ERKi will be most effective when used in combination, rather than as monotherapy. Although ERKi alone showed enhanced ability to suppress MAPK signaling in resistant clones relative to non-ERKi combinations (Fig. 2C), some degree of pathway output was still observed relative to control cells. This suggests that although ERKi may be more effective than MEKi in suppressing upstream signaling flux, ERKi may still be susceptible to some degree of adaptive feedback reactivation. Indeed, ERKi treatment led to dramatic feedback increases in pERK (Supplementary Fig. S2), but these effects were lessened in combination with BRAFi and to a greater extent with BRAFi/EGFRi. These findings are consistent with the modest degree of total cell growth observed with ERKi alone in the pooled clone system, despite lack of expansion in the relative abundance of resistant clones, suggesting that adaptive signaling changes affecting both control cells and resistant clones may have been responsible for this residual degree of proliferative potential. An additional rationale for using ERKi in combination with BRAFi and/or BRAFi/EGFRi in patients relates to therapeutic window. ERKi, like MEKi, inhibit MAPK signaling in tumor cells and normal cells, which may limit the effective dose achievable in patients due to on-target toxicity. In contrast, BRAFi inhibit MAPK signaling in cells with BRAFV600E, which can signal as a monomer, but do not inhibit MAPK signaling—and may even paradoxically activate the pathway—in cells with wild-type BRAF, where signaling occurs through RAF dimers, thus lessening the potential for on-target toxicity. Notably, although ERKi and BRAFi/ERKi led to enhanced suppression of cell growth and resistant clone expansion in vitro compared with BRAFi/MEKi/EGFRi (Figs. 2B and 3B–D), no significant difference in antitumor effect was noted in vivo between these groups, despite improved suppression of clonal outgrowth seen in the ERKi arms (Fig. 4A–C). It is possible that suboptimal in vivo dosing of ERKi may explain this difference. However, BRAFi/ERKi/EGFRi led to tumor regressions in vivo with complete suppression of clonal outgrowth (Fig. 4A–C) and an overall reduction in cell number and absolute clonal abundance in vitro (Fig. 3B–D). Overall, our data suggest that BRAFi/ERKi/EGFRi is the most effective therapy against the spectrum of clinically observed acquired resistance mechanisms, although the difference between BRAFi/ERKi/EGFRi and BRAFi/ERKi may be less pronounced in the subset of BRAFV600E colorectal cancers that do not rely on EGFR for feedback MAPK reactivation.
Acquired resistance is thought to arise most commonly from rare tumor subclones harboring preexisting resistance alterations. Indeed, Kopetz and colleagues (4) demonstrated that many BRAFV600E colorectal cancers harbor subclonal RAS mutations detectable at trace levels prior to treatment. Thus, in our study, we developed a pooled clone model in which multiple subclones, each harboring a distinct clinically observed resistance alteration, were spiked at low abundance (1% each) into a background of sensitive cells. In this way, we sought to generate a system that is more representative of the outgrowth of rare preexisting clones that occurs clinically, and that would allow us to test the ability of up-front convergent therapy strategies to suppress outgrowth of subclones bearing heterogeneous resistance alterations. Furthermore, we demonstrated that cfDNA can be isolated serially from culture media and that mutational abundance in cfDNA faithfully represents the clonal composition of the cultured cell population. Although cfDNA monitoring has proven a valuable tool in patients, this, to the best of our knowledge, is the first time cfDNA has been used in vitro, allowing us to monitor the kinetics of clonal outgrowth in real time during treatment. The ability to monitor in vitro clonal dynamics in real time provided key insights, including delayed kinetics of clonal outgrowth observed with BRAFi/MEKi/EGFRi—the most recent strategy tested in clinical trials for BRAFV600E colorectal cancer—and the ability of BRAFi/ERKi/EGFRi to lead to an overall reduction in absolute abundance of resistant subclones, indicative of cytotoxic effect. We propose that this pooled clone system can be a useful tool that can be readily applied to other acquired resistance paradigms to model outgrowth of heterogeneous resistance mechanisms under therapeutic pressure. Although only mutations were evaluated in this system using ddPCR, other important alterations, such as gene amplifications, could also be modeled through a modified approach, such as through utilization and monitoring of molecular barcodes. In fact, serial in vitro cfDNA monitoring may also be applied to gain kinetic insights in functional genomic studies using barcoded shRNA, ORF, or CRISPR libraries aimed at identifying novel resistance mechanisms or synthetic lethal interactions. Overall, these experiments provide important insight into acquired resistance and heterogeneity in BRAFV600E colorectal cancer and suggest that up-front therapy with BRAFi/ERKi or BRAFi/ERKi/EGFRi to suppress the rapid outgrowth of heterogeneous MAPK-altered resistant clones that limit the efficacy of current therapies for this disease may be a promising strategy for future clinical trials.
Detailed methods are included in Supplementary Materials.
Patient Samples, Cell Lines, and Reagents
Patient tumor and blood specimens were obtained from patients treated at the Massachusetts General Hospital under Institutional Review Board–approved studies and at the Dana-Farber Cancer Institute. All patients provided written, informed consent, and studies were conducted in accordance with the Declaration of Helsinki. Patients 2 and 3 were treated on clinical trials NCT01750918 and NCT01719380, respectively. Patients 1 and 4 were treated with dabrafenib 150 mg twice daily and panitumumab 6 mg/kg every 2 weeks alone or in combination with trametinib 2 mg daily, respectively, off-label with informed consent.
VACO432 cells were obtained from the Massachusetts General Hospital Center for Molecular Therapeutics, which performs routine cell line authentication testing by SNP and short-tandem repeat analysis in 2016 and were passaged less than 6 months following receipt. Cells were grown in DMEM/F12 (Gibco) with 10% FBS and assayed in DMEM/F12 with 5% FBS. Panitumumab was obtained from the McKesson Pharmaceuticals and diluted in PBS. Chemical inhibitors from the following sources were dissolved in DMSO: dabrafenib and trametinib (Selleck Chemicals) and VX-11e and ulixertinib (BVD-523; ChemieTek).
CT Scans, Whole-Exome Sequencing, and Targeted Sequencing
Spiral CT scans were obtained using standard procedures in the Department of Radiology at the Massachusetts General Hospital as part of the routine clinical care of these patients. Whole-exome sequencing of pretreatment and postprogression biopsies and normal blood was performed as previously described (patients 2 and 4; refs. 32–34). All BAM files were deposited in the database of Genotypes and Phenotypes (dbGaP), accession number phs000803.v2.p1. cfDNA was extracted from whole blood, and 5 to 30 ng of cfDNA was isolated. Sequencing libraries were prepared with custom in-line barcode molecular tagging, and complete sequencing at 15,000 × read depth of the critical exons in a targeted panel of 70 genes was performed at a Clinical Laboratory Improvement Amendments (CLIA)–certified, College of American Pathologists–accredited laboratory (Guardant Health; ref. 35). Targeted exome sequencing (patients 2 and 3) on clinical tissue specimens was performed as previously described (23).
cfDNA Isolation, Genome Equivalents Quantification (GE/mL Cell Culture Media) and ddPCR Analysis
For cell culture–derived cfDNA isolation, at least 2 mL of cell culture media was collected. Residual cells were removed through two rounds of centrifugation at 400 rcf for 5 minutes. cfDNA was extracted from purified cell culture media using the QIAmp Circulating Nucleic Acid Kit (Qiagen) according to the manufacturer's instructions. cfDNA (3 μL) was used as a template for each reaction. All samples were analyzed in triplicate. PCR reactions were performed using 10 μL final volume containing 5 μL LightCycler 480 SYBR Green I qPCR Master Mix, 2X (Roche) and LINE-1 (12.5 μmol/L) forward and reverse primers. DNA at known concentrations was also used to build the standard curve.
Isolated cfDNA and genomic DNA extracted from cells using the DNeasy Blood and Tissue Kit (Qiagen) was amplified using ddPCR Supermix for Probes (Bio-Rad) using NRAS, KRAS, MEK1, MEK2, and BRAF assays (PrimePCR ddPCR Mutation Assay, Bio-Rad, and custom-designed; Supplementary Table S3). DNA template (8 μL) was added to 10 μL of ddPCR Supermix for Probes (Bio-Rad) and 2 μL of the primer/probe mixture. This 20-μL reaction mix was added to a DG8 cartridge together with 70 μL of Droplet Generation Oil for Probes (Bio-Rad) and used for droplet generation. Droplets were then transferred to a 96-well plate (Eppendorf) and then thermal cycled with the following conditions: 5 minutes at 95°C, 40 cycles of 94°C for 30 seconds, 55°C for 1 minute followed by 98°C for 10 minutes (ramp rate 2°C/second). Droplets were analyzed with the QX200 Droplet Reader (Bio-Rad) for fluorescent measurement of FAM and HEX probes. Gating was performed based on positive and negative controls, and mutant populations were identified. The ddPCR data were analyzed with QuantaSoft analysis software (Bio-Rad) to obtain fractional abundance of the mutant DNA alleles in the wild-type (WT)/normal background. The quantification of the target molecule was presented as number of total copies (mutant plus WT) per sample in each reaction. Fractional abundance is calculated as follows: F.A. % = (Nmut/(Nmut + Nwt)) × 100), where Nmut is number of mutant events and Nwt is number of WT events per reaction. The number of positive and negative droplets is used to calculate the concentration of the target and reference DNA sequences and their Poisson-based 95% confidence intervals, as previously described (36). Multiple replicates (minimum of three) were performed for each sample. ddPCR analysis of normal control genomic DNA from cell lines and no DNA template (water) controls was performed in parallel with all samples, including multiple replicates as contamination-free controls.
Clonal pools were injected (5 × 106 cells per injection) into the flanks of 12-week-old female athymic nude mice (NU/NU; Charles River Laboratories). Genomic DNA was isolated in parallel from clonal pool for ddPCR analysis to confirm individual mutant peak levels. Once tumors reached an average volume of ∼100 to 200 mm3, mice were randomized into treatment arms, and tumor volume was assessed by caliper measurements over a 21-day period. The volume was calculated as L × W2 × π/6 and recorded twice per week. For ddPCR analysis, tumor tissue was harvested and snapped frozen. Dabrafenib and trametinib were formulated in 0.5% hydroxypropylmethylcellulose with 0.2% Tween 80 and dosed at 30 mg/kg and 1 mg/kg twice daily by oral gavage, respectively. Ulixertinib was formulated in 1% carboxymethylcellulose and dosed at 150 mg/kg twice daily by oral gavage. Panitumumab was formulated in PBS and dosed at 50 mg/kg dose twice weekly by intraperitoneal injection. Animal care and treatment was performed in accordance with institutional guidelines, and all experiments were conducted according to Institutional Animal Care and Use Committee–approved protocols.
Disclosure of Potential Conflicts of Interest
A.R. Parikh is a consultant/advisory board member for Eisai. E.L. Kwak is a clinical program leader at Novartis Biomedical Research Institutes. J.E. Faris is a clinical program leader at Novartis Biomedical Research Institutes and a consultant at N-of-One, and reports receiving commercial research support from Takeda, Roche, Exelixis, and Novartis. A.J. Iafrate has ownership interest (including patents) in ArcherDx and is a consultant/advisory board member for Chugai, DebioPharm, and Roche. E.M. Van Allen reports receiving commercial research grants from Novartis and BMS, has ownership interest (including patents) in Tango Therapeutics and Genome Medical, and is a consultant/advisory board member for Genome Medical and Invitae. R.B. Corcoran reports receiving commercial research grants from AstraZeneca and Sanofi and is a consultant/advisory board member for Amgen, Astex, Avidity Biosciences, BMS, Fog Pharma, Genentech, LOXO Oncology, Merrimack, N-of-One, Roche, Shire, and Taiho. No potential conflicts of interest were disclosed by the other authors.
Conception and design: M. Hazar-Rethinam, R.B. Corcoran
Development of methodology: M. Hazar-Rethinam, D. Liu, L.G. Ahronian, H.A. Shahzade, B. Nadres, E.M. Van Allen, R.B. Corcoran
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Hazar-Rethinam, M. Kleyman, H.A. Shahzade, L. Chen, J.N. Allen, J.W. Clark, E.L. Kwak, J.E. Faris, J.E. Murphy, T.S. Hong, E.E. Van Seventer, B. Nadres, C.B. Hong, J.M. Gurski Jr, N.A. Jessop, D. Dias-Santagata, A.J. Iafrate, R.B. Corcoran
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Hazar-Rethinam, M. Kleyman, G.C. Han, D. Liu, H.A. Shahzade, B. Nadres, D. Dias-Santagata, A.J. Iafrate, E.M. Van Allen, R.B. Corcoran
Writing, review, and/or revision of the manuscript: M. Hazar-Rethinam, M. Kleyman, D. Liu, A.R. Parikh, J.W. Clark, T.S. Hong, A.J. Iafrate, E.M. Van Allen, R.B. Corcoran
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L.G. Ahronian, N.A. Jessop, A.J. Iafrate, R.B. Corcoran
Study supervision: R.B. Corcoran
The work is supported by NIH/NCI Gastrointestinal Cancer SPORE P50 CA127003, R01CA208437, K08CA166510, U54CA224068, and a Damon Runyon Clinical Investigator Award. This research is supported by a Stand Up To Cancer Colorectal Cancer Dream Team Translational Research Grant (grant number: SU2C-AACR-DT22-17). Stand Up To Cancer (SU2C) is a program of the Entertainment Industry Foundation. Research grants are administered by the American Association for Cancer Research, the scientific partner of SU2C.